Rates of mental illness are much higher among people suffering from a substance use disorder (SUD) than in the rest of the population. A 10-year longitudinal follow-up of the National Comorbidity Survey confirmed that individuals with major depression at baseline had 40% higher odds of developing nicotine dependence and 90% higher odds of initiating the use of illicit drugs in the subsequent decade compared to individuals without major depression. Associations between depression and SUDs open the possibility that the successful treatment of pediatric depression could mitigate adverse consequences of SUDs, given an array of evidence-based treatments in pediatric populations and the fact that 70% of SUDs first emerge by the age of 20. However, rigorous evidence on whether depression treatment could mitigate SUDs and/or their consequences is scant, because it is difficult to ascertain if depression causes substance use, the reverse, or whether a third factor generates a spurious correlation between depression and SUDs. To examine how depression treatment may prevent or alter the onset and consequences of SUDs, we propose to take advantage of a powerful natural experiment: the sudden change in the prescribing patterns and use of antidepressants induced by the Food and Drug Administration's (FDA's) 2004 warnings on the risks of suicidality associated with pediatric antidepressant use. The warnings were extensively covered by the popular press, led to documented decreases in rates of identified depression, and reduced antidepressant use by 20 to 30% among pediatric patients with depression, making this a rare and powerful natural experiment. Furthermore, this abrupt change in prescribing patterns and use of antidepressants after the FDA warnings coincided with a 14% rise in youth suicide in the U.S. after 15 years of relatively steady decline in suicide rates, implying that the warnings profoundly affected depressive symptoms. Using three data sources that together provide information on substance use, abuse, and dependence, medical care utilization (including data on prescription drugs), and academic and crime related outcomes, the proposed research aims to: 1) Examine whether FDA warnings regarding pediatric antidepressant use and the concomitant drop in pediatric antidepressant use increased the onset or intensity of substance use, abuse, and dependence (for tobacco, alcohol, illicit drugs, and illicit use of prescription drugs); 2) Examine whether FDA warnings regarding pediatric antidepressant use and the concomitant drop in pediatric antidepressant use increased health care use and costs; and 3) Examine whether FDA warnings regarding pediatric antidepressant use and the concomitant drop in pediatric antidepressant use increased the social costs of substance use, abuse, and dependence (health care use, poor school performance, delinquency, criminal activity). Our analysis will use a regression discontinuity design to compare the prescribing patterns and SUD outcomes of youths with new onset of depression just before and immediately after the 2004 FDA warnings on pediatric antidepressant use.